Background
Neurodegenerative diseases are chronic and inexorable conditions characterized by the presence of insoluble aggregates of abnormally ubiquitinated and phosphorylated proteins [
1]. Transactive response DNA-binding protein 43 (TDP-43) is one such protein implicated in the neurodegenerative process and the presence of TDP-43 inclusions in neurons is a hallmark finding in amyotrophic lateral sclerosis (ALS) [
2,
3]. Recent evidence also points to a prion-like self-propagation of TDP-43 misfolding, either by the circulatory system, cell-to-cell contact, or via the interstitial or cerebrospinal fluids (CSF) [
1,
4‐
6]. As protein aggregation occurs before the onset of brain damage and motor symptoms, new therapeutic strategies targeting the spread of disease across the brain (including eliminating seed proteins and blocking cell-to-cell spread) are of vital importance.
One potential therapeutic target is the glymphatic system, a CSF-based waste clearance system for the brain [
7]. Mediated by a unique system of astrocyte-specific aquaporin-4 water channels, the glymphatic system is largely dormant during wakefulness but highly active during sleep, working to clear waste byproducts from the brain through the flow of CSF [
7]. Preclinical studies have shown that amyloid-β, a protein implicated in neurodegenerative diseases including Alzheimer’s disease, is cleared from the brain by the glymphatic system [
7] and that in aged mice, amyloid-β clearance is dramatically slowed [
8]. As the prevalence of ALS also increases with age, sleep disturbances are exceedingly common in ALS [
9] and given that a single night of sleep deprivation can result in amyloid-β accumulation linked with Alzheimer’s disease [
10], we hypothesize that glymphatic clearance is also impaired in ALS.
To investigate this, we assessed glymphatic function in a TDP-43 transgenic mouse model of ALS. Specifically, these mice have doxycycline (Dox)-suppressible neurofilament heavy chain (
NEFH) promoter-driven expression of human TDP-43 (hTDP-43) harboring a defective nuclear localization signal (hTDP-43ΔNLS). When these mice are removed from Dox feed, hTDP-43 becomes expressed, resulting in the accumulation of insoluble phosphorylated cytoplasmic TDP-43 in neurons of the brain and spinal cord and loss of endogenous nuclear mouse TDP-43 [
11]. Continued hTDP-43 expression has been shown to result in much of the pathophysiology seen in ALS, including brain atrophy, cortical and spinal motor neuron loss, muscle denervation, and progressive motor impairments, leading to premature death.
We performed longitudinal multimodal magnetic resonance imaging (MRI) one and three weeks after the cessation of Dox feed, together with weekly rotarod assessments of motor performance. Glymphatic function was assessed using dynamic contrast-enhanced MRI (DCE-MRI) to track the clearance of an MR contrast agent injected into the cisterna magna [
12]. DCE-MRI was performed following the second multimodal imaging session three weeks after the cessation of Dox feed.
Telomere analysis
DNA extraction from 25-mg ear tissue was performed using DNeasy® Blood and Tissue kit (QIAGEN, Hilden, Germany) according to the manufacturer’s instructions. The quality and quantity of DNA was measured using the QIAxpert system and samples with an A260/A280 ratio of 1.7–2 were used. DNA was diluted to 20 ng/µl in Tris–EDTA buffer solution for further analysis. Two master mixes containing telomere repeat copy and a single copy gene (36B4) were used. The qPCR mix contained 1 µl DNA, 1× SYBR Green master mix (Promega, Madison, WI), and forward and reverse primers. The final primer concentrations were as follows: telomere forward 270 nM, telomere reverse 900 nM, 36B4 forward 300 nM and 36B4 reverse 500 nM.
Auto-aliquoting was performed using the QIAgility liquid handling robot and samples were run in duplicates with primer sequences as described previously [
16]. The thermocycling conditions for telomere were 1 cycle at 95 °C for 3 min, 30 cycles of 95 °C for 15 s and 54.8 °C for 1 min followed by the melting curve, and for 36B4 were 1 cycle at 95 °C for 3 min, 30 cycles of 95 °C for 15 s and 58.4 °C for 1 min followed by the melting curve. Telomere length was determined by comparing telomere to 36B4 as described previously [
17].
Statistical analyses
Voxel-wise analysis of tensor-based morphometry was performed using the FSL’s Randomise with threshold-free cluster enhancement [
18], and results were fully corrected for multiple comparisons. Changes in ΔFDC were assessed using connectivity-based fixel enhancement with results fully corrected for family-wise error (FWE). Corticospinal tract analyses of FDC were tested using a general linear model with time as repeated measure (1 and 3 weeks post-Dox) and genotype (TDP-43, control) and position along the tracts (p1 to p20) as factors. Statistical testing was performed using the IBM SPSS Statistics for Macintosh, Version 27.0.
Two-way ANOVAs or mixed-effects analyses with genotype (TDP-43 vs control) as the between-subject factor and time (1 week vs 3 weeks post-Dox) as the within-subject factor were used to assess behavior and ROI-based measures. Šídák’s multiple comparison tests were performed where necessary. Data on telomere length were analyzed with an unpaired t-test. Statistical testing for these comparisons was performed using GraphPad Prism version 9.1.0 software (GraphPad, San Diego, CA). Significance for all tests was set at P < 0.05 and all results are presented as mean ± standard error of the mean (SEM).
Discussion
In this study, we assessed glymphatic function in a transgenic mouse model with Dox-suppressible expression of hTDP-43. The mice were removed from Dox feed to permit hTDP-43 expression, then they were assessed longitudinally using rotarod testing and underwent in vivo multimodal MRI followed by DCE-MRI at 3 weeks post-Dox cessation. Our results demonstrate that the TDP-43 mice exhibited significantly altered glymphatic function, progressive neurodegeneration, deteriorating motor symptoms, significant weight loss, and shortened telomere length when compared to their littermate controls, very early in the disease course.
Waste protein inclusions are a hallmark of all neurodegenerative diseases and recent efforts have recognized the importance of the glymphatic system for protein regulation and brain clearance. This brain-wide network of perivascular space facilitates the exchange of CSF and interstitial fluid, clearing waste from the CSF while also assisting in distributing compounds critical to neurological function. Here, we used DCE-MRI to investigate glymphatic function following the injection of an MR contrast agent, Magnevist, into the cisterna magna, a technique first demonstrated by Iliff and colleagues [
12].
As hypothesized, glymphatic function was found to be altered in TDP-43 mice at 3 weeks post-Dox cessation, a time point prior to overt neurodegeneration (4 weeks) and very early in the disease course in this model [
11]. Although this is the first in vivo study of glymphatic function in ALS and one of the few DCE-MRI studies conducted in mice, similar results demonstrating increased signal with impaired glymphatic function have been observed in rodents previously [
26,
27]. One recent study employed DCE-MRI to investigate the effect of the circadian light/dark cycle in awake rats [
28]. This study also showed similar brain-wide kinetics, with increased signal during the dark (awake) phase—when glymphatic clearance is known to be reduced. Although we did not acquire images during contrast agent infusion and hence were unable to similarly quantify the time to peak signal increase, we demonstrated significant interactions between time and genotype for each of the regions analyzed, and further, found a delay in contrast agent uptake in the thalamus, suggesting impaired glymphatic function in TDP-43 mice.
Our preliminary results highlight the need for additional experiments, at multiple time points, to improve our understanding of the relationship between ALS and glymphatic system function, and how this relationship evolves with disease progression. In addition to measuring signal change during contrast agent infusion for the spatial quantification of measures such as peak signal increase and the time to peak signal increase, future experiments should also consider the use of α
2 agonists such as xylazine and dexmedetomidine [
29,
30]. When compared to isoflurane alone, dexmedetomidine and low-dose isoflurane significantly increased glymphatic transport [
30] and the interstitial fluid space [
31] and hence, may better mimic sleep [
32].
In addition to DCE-MRI, we performed in vivo MRI at 1 and 3 weeks after the cessation of Dox feed. Structural analysis of hippocampal volume demonstrated an interaction between genotype and time, with volumes increasing in control mice and decreasing over time in TDP-43 mice. Further, tensor-based morphometry also revealed atrophy of the hippocampus at 3 weeks post-Dox cessation. Altogether, these results suggest a possible loss of hippocampus neurons in TDP-43 mice over time. Hippocampal pathology is well documented in ALS patients primarily at later stages of disease [
3,
20,
33‐
35] and a recent experimental study also demonstrated significant neuronal loss and atrophy of the hippocampus, after intracranial injection of recombinant adeno-associated virus serotype 9 containing human wild-type TDP-43, into the hippocampus of
CAMKII-tTa transgenic mice [
36]. Expression of a C-terminal fragment of TDP-43 found in the brains of frontotemporal lobar degeneration cases, has also been shown to cause a specific loss of hippocampal dentate gyrus neurons in mice, although the relevance for this finding to ALS remains unclear [
37].
DWI has demonstrated potential as a sensitive biomarker for neurodegeneration, identifying microstructural changes to the white matter in the absence of any macroscopic changes on conventional radiological images [
22,
23]. We assessed the fixel-derived metric FDC using both exploratory whole-brain and a priori analyses. As the name suggests, changes in the combined fibre density and cross-section measure reflect both a macroscopic change to the white matter bundle and a microstructural change due to increased or decreased axon fibre population [
15,
38]. Further, as a fixel-derived metric, FDC can be assessed for specific fibre bundle orientations and hence, specific white matter tracts of interest.
The corticospinal tract is the major neuronal pathway controlling movement and has previously been implicated in DWI studies of ALS [
13,
24,
25,
39]. Using connectivity-based fixel analysis, a whole-brain exploratory method, we found significantly reduced FDC in the left corticospinal tract of TDP-43 mice over time. Additionally, hypothesis-driven tract-of-interest analyses of bilateral corticospinal tract FDC values also demonstrated significant reductions in TDP-43 mice compared to littermate controls. Post-hoc comparisons demonstrated that the TDP-43 mice had significantly reduced FDC in the corticospinal tracts at 1 week post-Dox cessation, largely at the caudal end, near the cerebral peduncles. These changes were observed in the absence of any structural change to the primary motor cortices on T
2*-weighted imaging, consistent with the hypothesis that axonopathy may precede the degeneration of neuronal cell bodies—i.e. ‘dying-back pathology’ [
19,
40‐
42].
Consistent with the progressive corticospinal tract degeneration and the ALS phenotype, the TDP-43 mice also demonstrated progressive weight loss and deteriorating motor deficits when compared to controls. The motor performance of TDP-43 mice was significantly impaired compared to controls at 1 week post-Dox cessation, with over half also demonstrating a hind-limb clasping phenotype. At 3 weeks post-Dox cessation, clasping was evident in all TDP-43 mice and motor performance had further declined. Interestingly, the rotarod latency to fall correlated significantly with the mean caudal corticospinal tract FDC values, suggesting that the decline of motor function is closely associated with corticospinal tract degeneration.
As expected, the gastrocnemii weight of TDP-43 mice was significantly lighter than that of littermate controls. Post-mortem analysis also demonstrated a trend of smaller spleens in these mice, consistent with observations of reduced spleen weight late in the disease course in SOD1
G93A mice [
43,
44]. Peripheral immune dysfunction is a pathogenic feature of clinical ALS, with altered levels of T-lymphocytes, monocytes and cytokines in the blood [
45], and it is possible that similar immune system changes may occur in this TDP-43 mouse model.
Chronic systemic inflammation and immune cell exhaustion are also hypothesized sources of telomere shortening [
46]. Telomeres are special chromatin structures formed at the ends of chromosomes, protecting them from degradation and recombination [
47]. Each time a cell divides, the telomeres shorten, and hence telomere length is considered a measure of ageing [
47]. Telomere shortening has been reported in other neurodegenerative diseases, including traumatic brain injury [
17,
48,
49] and Alzheimer’s disease [
50,
51]. Consistent with this, we found that telomere length was significantly shorter in TDP-43 mice when compared to controls.
In both sporadic ALS patients and healthy controls, telomere length has been shown to correlate with decreased telomerase reverse transcriptase expression [
52]. In addition to maintaining telomere length, telomerase may also protect against cell damage. Increasing telomerase expression has been shown to delay disease onset in SOD1
G93A mice [
53], while conversely, telomerase deletion accelerates disease progression in SOD1
G93A mice [
54]. These results, together with our observation of reduced telomere length in TDP-43 mice, suggest telomere/telomerase dysfunction in the pathogenesis of ALS.